Arnica and a scientist’s frustrations

When you’re the only scientist in the family you get asked the weirdest things. Actually, I’m not the only one, but the other one is a chemist and he’s mostly asked about astrophysics stuff, so he doesn’t really count, because I am the one who gets asked about rare diseases and medication side-effects and food advice. Never mind that I am a neuroscientist and I have professed repeatedly and quite loudly my minimum knowledge of everything from the neck down, all eyes turn to me when the new arthritis medication or the unexpected side-effects of that heart drug are being brought up. But, curiously, if I dare speak about brain stuff I get the looks that a thing the cat just dragged in gets. I guess everybody is an expert on how the brain works on account of having and using one, apparently. Everybody, but the actual neuroscience expert whose input on brain and behavior is to be tolerated and taken with a grain of salt at best, but whose opinion on stomach distress is of the utmost importance and must be listened to reverentially in utter silence [eyes roll].

So this is the background on which the following question was sprung on me: “Is arnica good for eczema?”. As always, being caught unawares by the sheer diversity of interests and afflictions my family and friends can have, I mumbled something about I don’t know what arnica is and said I will look it up.

This is an account of how I looked it up and what conclusions I arrived to or how a scientist tries to figure out something completely out of his or her field. First thing I did was to go on Wikipedia. Hold your horses, it was not about scientific information but for a first clarification step: is it a chemical, a drug, an insect, a plant maybe? I used to encourage my students to also use Wikipedia when they don’t have a clue what a word/concept/thing is. Kind of like a dictionary or a paper encyclopedia, if you will. To have a starting point. As a matter of fact Wikipedia is an online encyclopedia, right? Anyway, I found out that Arnica is a plant genus out of which one species, Arnica Montana, seems to be popular.

Then I went to the library. Luckily for me, the library can be accessed online from the comfort of my home and in my favorite pajamas in the incarnation of PubMed or Medline as it used to be affectionately called. It is the US National Library of Medicine maintained by the National Institutes of Health, a wonderful repository of scholarly papers (yeah, Google Scholar to PubMed is like the babbling of a two-year old to the Shakespearian sonnets; Google also has an agenda, which you won’t find on PubMed). Useful tip: when you look for a paper that is behind a paywall in Nature or Elsevier Journals or elsewhere, check the PubMed too because very few people seem to know that there is an obscure and incredibly helpful law saying that research paid by the US taxpayers should be available to the US taxpayer. A very sensible law passed only a few years ago that has the delightful effect of having FREE full text access to papers after a certain amount of months from publishing (look for the PMC icon in the upper right corner).

I searched for “arnica” and got almost 400 results. I sorted by “most recent”. The third hit was a review. I skimmed it and seemed to talk a lot about healing in homeopathy, at which point, naturally, I got a gloomy foreboding. But I persevered because one data point does not a trend make. Meaning that you need more than a paper – or a handful – to form an informed opinion. This line of thinking has been rewarded by the hit No. 14 in the search which had an interesting title in the sense that it was the first to hint to a mechanism through which this plant was having some effects. Mechanisms are important, they allow you to differentiate speculation from findings, so I always prefer papers that try to answer a “How?” question as opposed to the other kinds; whys are almost always speculative as they have a whiff of post factum rationalizations, whats are curious observations but, more often than not, a myriad factors can account for them, whens are an interesting hybrid between the whats and the hows – all interesting reads but for different purposes. Here is a hint: you want to publish in Nature or Science? Design an experiment that answers all the questions. Gone are the days when answering one question was enough to publish…

Digressions aside, the paper I am covering today sounds like a mechanism paper. Marzotto et al. (2016) cultured a particular line of human cells in a Petri dish destined to test the healing powers of Arnica montana. The experimental design seems simple enough: the control culture gets nothing and the experimental culture gets Arnica montana. Then, the authors check to see if there are differences in gene expressions between the two groups.

The authors applied different doses of Arnica montana to the cultures to see if the effects are dose-dependant. The doses used were… wait, bear with me, I’m not familiar with the system, it’s not metric. In the Methods, the authors say

“Arnica m. was produced by Boiron Laboratoires (Lyon, France) according to the French Homeopathic pharmacopoeia and provided as a first centesimal dilution (Arnica m. 1c) of the hydroalcoholic extract (Mother Tincture, MT) in 30% ethanol/distilled water”.

Wait, what?! Centesimal… centesimal… wasn’t that the nothing-in-it scale from the pseudoscientific bull called homeopathy? Maybe I’m wrong, maybe there are some other uses for it and becomes clear later:

“Arnica m. 1c was used to prepare the second centesimal dilution (Arnica m. 2c) by adding 50μl of 1c solution to 4.95ml of distilled ultra-pure water. Therefore, 2c corresponds to 10−4 of the MT”.

Holy Mother of God, this is worse than gibberish; this is voluntary misdirection, crap wrapped up in glitter, medieval tinkering sold as state-of-the-art 21st century science. Speaking of state-of-the-art, the authors submit their “doses” to a liquid chromatograph, a thin layer chromatograph, a double-beam spectrophotometer, a nanoparticle tracking analysis (?!) for what purposes I cannot fathom. On, no, I can: to sound science-y. To give credibility for the incredulous. To make money.

At which point I stopped reading the ridiculous nonsense and took a closer look at the authors and got hit with this:

“Competing Interests: The authors have declared that no competing interests exist. This study was funded by Boiron Laboratoires Lyon with a research agreement in partnership with University of Verona. There are no patents, products in development or marketed products to declare. This does not alter our adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.”

No competing interests?? The biggest manufacturer of homeopathic crap in the world pays you to see if their product works and you have no competing interest? Maybe no other competing interests. There were some comments and replies to this paper after that, but it is all inconsequential because once you have faulty methods your results are irrelevant. Besides, the comments are from the same University, could be some internal feuding.

PLoS One, what have you done? You’re a peer-reviewed open access journal! What “peers” reviewed this paper and gave their ok for publication? Since when is homeopathy science?! What am I going to find that you publish next? Astrology? For shame… Give me that editor’s job because I am certain I can do better.

To wrap it up and tell you why I am so mad. The homeopathic scale system, that centesimal gibberish, is just that: gibberish. It is impossible to replicate this experiment without the product marketed by Boiron because nobody knows how much of the plant is in the dose, which parts of the plant, what kind of extract, or what concentration. So it’s like me handing you my special potion and telling you it makes warts disappear because it has parsley in it. But I don’t tell you my recipe, how much, if there anything else besides parsley in it, if I used the roots or only the leaves or anything. Now that, my friends, it’s not science, because science is REPLICABLE. Make no mistake: homeopathy is not science. Just like the rest of alternative medicine, homeopathy is a ruthless and dangerous business that is in sore need of lawmakers’ attention, like FDA or USDA. And for those who think this is a small paper, totally harmless, no impact, let me tell you that this paper had over 20,000 views (real science papers get hundreds, maybe thousands).

I would have oh so much more to rant on. But enough. Rant over.

Oh, not yet. Lastly, I checked a few other papers about arnica and my answer to the eczema question is: “It’s possible but no, I don’t think so. I don’t know really, I couldn’t find any serious study about it and I gave up looking after I found a lot of homeopathic red flags”. The answer I will give my family member? “Not the product you have, no. Go to the doctors, the ones with MDs after their name and do what they tell you. In addition, I, the one with a PhD after my name, will tell you this for free because you’re family: rub the contents of this bottle only once a day – no more! – on the affected area and you will start seeing improvements in three days. Do not use elsewhere, it’s quite potent!” Because placebo works and at least my water vial is poison free.

Reference: Marzotto M, Bonafini C, Olioso D, Baruzzi A, Bettinetti L, Di Leva F, Galbiati E, & Bellavite P (10 Nov 2016). Arnica montana Stimulates Extracellular Matrix Gene Expression in a Macrophage Cell Line Differentiated to Wound-Healing Phenotype. PLoS One, 11(11):e0166340. PMID: 27832158, PMCID: PMC5104438, DOI: 10.1371/journal.pone.0166340. ABSTRACT | FREE FULLTEXT PDF 

By Neuronicus, 10 June 2017

EDIT [7 December 2022]: Look at that! There is a warning after the DOI that appeared on my website through no involvement of my own saying the paper has been retracted roughly 2 years after publication. Good! The official reason for retraction? They used nothing to find nothing, their “findings” are just “standard noise”. Told you so! PubPeer link here.

This is what PLoS One is saying about it (my highlights):

The PLOS ONE Editors have discussed the study design and results reported in this article with experts in RNA-seq analysis, statistical analysis and members of our Editorial Board. Based on our assessment and the advice received, and in light of the above concerns, we have determined that the results presented in this article do not provide sufficient support for claims about effects of Arnica m. on gene expression. Hence, we are retracting this article due to concerns about the study design and about the validity and reliability of the reported conclusions. We regret that these issues were not fully addressed during the article’s pre-publication peer review.
In addition to the above, the PLOS ONE Editors hereby notify readers that the Competing Interests statement was incorrect for this article and should have explicitly stated that Boiron Laboratories, a company that provided funding support for this study, markets homeopathic products including various dilutions of Arnica m.
MM, CB, DO, EG, and PB did not agree with retraction. AB, LB, and FDL did not respond.

Painful Pain Paper

There has been much hype over the new paper published in the latest Nature issue which claims to have discovered an opioid analgesic that doesn’t have most of the side effects of morphine. If the claim holds, the authors may have found the Holy Grail of pain research chased by too many for too long (besides being worth billions of dollars to its discoverers).

The drug, called PZM21, was discovered using structure-based drug design. This means that instead of taking a drug that works, say morphine, and then tweaking its molecular structure in various ways and see if the resultant drugs work, you take the target of the drug, say mu-opioid receptors, and design a drug that fits in that slot. The search and design are done initially with sophisticated software and there are many millions of virtual candidates. So it takes a lot of work and ingenuity to select but a few drugs that will be synthesized and tested in live animals.

Manglik et al. (2016) did just that and they came up with PZM21 which, compared to morphine, is:

1) selective for the mu-opioid receptors (i.e. it doesn’t bind to anything else)
2) produces no respiratory depression (maybe a touch on the opposite side)
3) doesn’t affect locomotion
4) produces less constipation
5) produces long-lasting affective analgesia
6) and has less addictive liability

The Holy Grail, right? Weeell, I have some serious issues with number 5 and, to some extent, number 6 on this list.

Normally, I wouldn’t dissect a paper so thoroughly because, if there is one thing I learned by the end of GradSchool and PostDoc, is that there is no perfect paper out there. Consequently, anyone with scientific training can find issues with absolutely anything published. I once challenged someone to bring me any loved and cherished paper and I would tear it apart; it’s much easier to criticize than to come up with solutions. Probably that’s why everybody hates Reviewer No. 2…

But, for extraordinary claims, you need extraordinary evidence. And the evidence simply does not support the 5 and maybe 6 above.

Let’s start with pain. The authors used 3 tests: hotplate (drop a mouse on a hot plate for 10 sec and see what it does), tail-flick (give an electric shock to the tail and see how fast the mouse flicks its tail) and formalin (inject an inflammatory painful substance in the mouse paw and see what the animal does). They used 3 doses of PZM21 in the hotplate test (10, 20, and 40 mg/Kg), 2 doses in the tail-flick test (10 and 20), and 1 dose in the formalin test (20). Why? If you start with a dose-response in a test and want to convince me it works in the other tests, then do a dose-response for those too, so I have something to compare. These tests have been extensively used in pain research and the standard drug used is morphine. Therefore, the literature is clear on how different doses of morphine work in these tests. I need your dose-responses for your new drug to be able to see how it measures up, since you claim it is “more efficacious than morphine”. If you don’t want to convince me there is a dose-response effect, that’s fine too, I’ll frown a little, but it’s your choice. However, then choose a dose and stick with it! Otherwise I cannot compare the behaviors across tests, rendering one or the other test meaningless. If you’re wondering, they used only one dose of morphine in all the tests, except the hotplate, where they used two.

Another thing also related to doses. The authors found something really odd: PZM21 works (meaning produces analgesia) in the hotplate, but not the tail-flick tests. This is truly amazing because no opiate I know of can make such a clear-cut distinction between those two tests. Buuuuut, and here is a big ‘BUT” they did not test their highest dose (40mg/kg) in the tail-flick test! Why? I’ll tell you how, because I am oh sooo familiar with this argument. It goes like this:

Reviewer: Why didn’t you use the same doses in all your 3 pain tests?

Author: The middle and highest doses have similar effects in the hotplate test, ok? So it doesn’t matter which one of these doses I’ll use in the tail-flick test.

Reviewer: Yeah, right, but, you have no proof that the effects of the two doses are indistinguishable because you don’t report any stats on them! Besides, even so, that argument applies only when a) you have ceiling effects (not the case here, your morphine hit it, at any rate) and b) the drug has the expected effects on both tests and thus you have some logical rationale behind it. Which is not the case here, again: your point is that the drug DOESN’T produce analgesia in the tail-flick test and yet you don’t wanna try its HIGHEST dose… REJECT AND RESUBMIT! Awesome drug discovery, by the way!

So how come the paper passed the reviewers?! Perhaps the fact that two of the reviewers are long term publishing co-authors from the same University had something to do with it, you know, same views predisposes them to the same biases and so on… But can you do that? I mean, have reviewers for Nature from the same department for the same paper?

Alrighty then… let’s move on to the stats. Or rather not. Because there aren’t any for the hotplate or tail-flick! Now, I know all about the “freedom from the tyranny of p” movement (that is: report only the means, standard errors of mean, and confidence intervals and let the reader judge the data) and about the fact that the average scientist today needs to know 100-fold more stats that his predecessors 20 years ago (although some biologists and chemists seem to be excused from this, things either turn color or not, either are there or not etc.) or about the fact that you cannot get away with only one experiment published these days, but you need a lot of them so you have to do a lot of corrections to your stats so you don’t fall into the Type 1 error. I know all about that, but just like the case with the doses, choose one way or another and stick to it. Because there are ANOVAs ran for the formalin test, the respiration, constipation, locomotion, and conditioned place preference tests, but none for the hotplate or tailflick! I am also aware that to be published in Science or Nature you have to strip your work and wordings to the bare minimum because the insane wordcount limits, but you have free rein in the Supplementals. And I combed through those and there are no stats there either. Nor are there any power analyses… So, what’s going on here? Remember, the authors didn’t test the highest dose on the tail-flick test because – presumably – the highest and intermediary doses have indistinguishable effects, but where is the stats to prove it?

And now the thing that really really bothered me: the claim that PZM21 takes away the affective dimension of pain but not the sensory. Pain is a complex experience that, depending on your favourite pain researcher, has at least 2 dimensions: the sensory (also called ‘reflexive’ because it is the immediate response to the noxious stimulation that makes you retract by reflex the limb from whatever produces the tissue damage) and the affective (also called ‘motivational’ because it makes the pain unpleasant and motivates you to get away from whatever caused it and seek alleviation and recovery). The first aspect of pain, the sensory, is relatively easy to measure, since you look at the limb withdrawal (or tail, in the case of animals with prolonged spinal column). By contrast, the affective aspect is very hard to measure. In humans, you can ask them how unpleasant it is (and even those reports are unreliable), but how do you do it with animals? Well, you go back to humans and see what they do. Humans scream “Ouch!” or swear when they get hurt (so you can measure vocalizations in animals) or humans avoid places in which they got hurt because they remember the unpleasant pain (so you do a test called Conditioned Place Avoidance for animals, although if you got a drug that shows positive results in this test, like morphine, you don’t know if you blocked the memory of unpleasantness or the feeling of unpleasantness itself, but that’s a different can of worms). The authors did not use any of these tests, yet they claim that PZM21 takes away the unpleasantness of pain, i.e. is an affective analgesic!

What they did was this: they looked at the behaviors the animal did on the hotplate and divided them in two categories: reflexive (the lifting of the paw) and affective (the licking of the paw and the jumping). Now, there are several issues with this dichotomy, I’m not even going to go there; I’ll just say that there are prominent pain researchers that will scream from the top of their lungs that the so-called affective behaviors from the hotplate test cannot be indexes of pain affect, because the pain affect requires forebrain structures and yet these behaviors persist in the decerebrated rodent, including the jumping. Anyway, leaving the theoretical debate about what those behaviors they measured really mean aside, there still is the problem of the jumpers: namely, the authors excluded from the analysis the mice who tried to jump out of the hotplate test in the evaluation of the potency of PZM21, but then they left them in when comparing the two types of analgesia because it’s a sign of escaping, an emotionally-valenced behavior! Isn’t this the same test?! Seriously? Why are you using two different groups of mice and leaving the impression that is only one? And oh, yeah, they used only the middle dose for the affective evaluation, when they used all three doses for potency…. And I’m not even gonna ask why they used the highest dose in the formalin test… but only for the normal mice, the knockouts in the same test got the middle dose! So we’re back comparing pears with apples again!

Next (and last, I promise, this rant is way too long already), the non-addictive claim. The authors used the Conditioned Place Paradigm, an old and reliable method to test drug likeability. The idea is that you have a box with 2 chambers, X and Y. Give the animal saline in chamber X and let it stay there for some time. Next day, you give the animal the drug and confine it in chamber Y. Do this a few times and on the test day you let the animal explore both chambers. If it stays more in chamber Y then it liked the drug, much like humans behave by seeking a place in which they felt good and avoiding places in which they felt bad. All well and good, only that is standard practice in this test to counter-balance the days and the chambers! I don’t know about the chambers, because they don’t say, but the days were not counterbalanced. I know, it’s a petty little thing for me to bring that up, but remember the saying about extraordinary claims… so I expect flawless methods. I would have also liked to see a way more convincing test for addictive liability like self-administration, but that will be done later, if the drug holds, I hope. Thankfully, unlike the affective analgesia claims, the authors have been more restrained in their verbiage about addiction, much to their credit (and I have a nasty suspicion as to why).

I do sincerely think the drug shows decent promise as a painkiller. Kudos for discovering it! But, seriously, fellows, the behavioral portion of the paper could use some improvements.

Ok, rant over.

EDIT (Aug 25, 2016): I forgot to mention something, and that is the competing financial interests declared for this paper: some of its authors already filed a provisional patent for PZM21 or are already founders or consultants for Epiodyne (a company that that wants to develop novel analgesics). Normally, that wouldn’t worry me unduly, people are allowed to make a buck from their discoveries (although is billions in this case and we can get into that capitalism-old debate whether is moral to make billions on the suffering of other people, but that’s a different story). Anyway, combine the financial interests with the poor behavioral tests and you get a very shoddy thing indeed.

Reference: Manglik A, Lin H, Aryal DK, McCorvy JD, Dengler D, Corder G, Levit A, Kling RC, Bernat V, Hübner H, Huang XP, Sassano MF, Giguère PM, Löber S, Da Duan, Scherrer G, Kobilka BK, Gmeiner P, Roth BL, & Shoichet BK (Epub 17 Aug 2016). Structure-based discovery of opioid analgesics with reduced side effects. Nature, 1-6. PMID: 27533032, DOI: 10.1038/nature19112. ARTICLE 

By Neuronicus, 21 August 2016

Orgasm-inducing mushrooms? Not quite

Claims that there is an orgasm-inducing mushroom in Hawaii may not be entirely accurate. Author and licensing of the above drawing unknown.

A few weeks ago, the social media has bombarded us with the eye-catching news that there is a mushroom in Hawaii whose smell induces spontaneous orgasms in women, but not men, who found its smell repugnant.

Except that it appears there is no such mushroom. Turns out the 14 year old paper is written by the president of a Hawaiian company that sells organic medicinal mushrooms. Not only written, but funded, as well. This is enough to damn the credibility of any study (that’s why scientists must declare competing interest when submitting a paper). But it also seems that the study has major fundamental flaws, like not having a single objective measure (of the quantity of spores, for example), is done under non-controlled environmental conditions (the participants seem to have known what was expected from them), there have been no replications, etc. Actually, it should have been suspicious to me from the start that nothing happened in the following 14 years; you would think that such claims would have been replicated, or at least the mushroom identified. But, as they say, hindsight is 20-20. Here is some nice little reporting exposing the business in Huffington Post and ScienceAlert.

I am not blaming the science media outlets on this one too much, like IFL Science or NBC affiliate, as I thought of covering this study myself, should I have been able to get my hands on the full text of the paper. In all honesty, who wouldn’t want to read that paper, especially since the abstract speculated on the mushroom’s spores having hormone-like chemicals that mimic the human neurotransmitters released during sexual encounters? But I (and others) have searched in vain for the full text and the most parsimonious explanation is that it was buried or withdrawn.

The trite but true message is: even the science media (including this one) is prone to mistakes. Interested in something? Go to the source and read the whole paper yourself, even the small print (like the one with competing interests), and only then make an opinion. That’s why I always post the links to the original article.

Reference: Holliday, J.C. & Soule, N. (2001). Spontaneous Female Orgasms Triggered by Smell of a Newly Found Tropical Dictyphora Species. International Journal of Medicinal Mushrooms, 3: 162-167. Abstract | Debunking in The Journal of Wild Mushrooming | Debunking in Discover Magazine

By Neuronicus, 17 October 2015